Mechanical Loading Histories and Cortical Bone Remodeling. Dennis R. Carter. Design Division, Department of Mechanical Engineering, Stanford University, ...
Calcif Tissue lnt (1984) 36:S19-$24
Calcified Tissue International 9 1984 by Springer-Vedag
Mechanical Loading Histories and Cortical Bone Remodeling Dennis R. Carter Design Division, Department of Mechanical Engineering, Stanford University, Stanford, CA 94305; and Rehabilitation Research and Development Center, Palo Alto Veterans Administration Medical Center, Palo Alto, CA 94304, USA
Summary. A conceptual framework is presented for understanding and investigating structural adaptation of cortical bone. The magnitudes, orientations, and sense (tension or compression) of the physiologically incurred cyclic principal strains vary markedly throughout the skeleton. It is probable, therefore, that the strain/remodeling response of bone is site specific. Furthermore, there is some indication that immature bone is more responsive to alterations of cyclic strains than mature bone. Animal experimental studies and complementary stress and strain analyses suggest that the structural adaptation due to changes in cyclic strain fields may be a very nonlinear response. Bone loss in mature animals due to immobilization is sensitive to even small changes in the cyclic bone strains. Under normal conditions, however, there appears to be a broad range of physical activity in which bone is relatively unresponsive to changes in loading history. With severe repeated loading, bone hypertrophy can be pronounced. These observations open the possibility that bone atrophy and hypertrophy are controlled by different mechanisms. Therefore, two (or more) complementary control systems may be involved in the regulation of bone mass by bone cyclic strain histories. It is probable that bone mechanical microdamage is one control stimulus for affecting an increase in bone mass. Key words: Bone - - Remodeling - - Microdamage.
In the 17th century, Galileo inferred that a relationship exists between mechanical forces and bone morphology when he noted that body weight and
Supported by NIH grants AM32377 and AM01163 (RCDA) Send offprint requests to D. R. Carter at the Stanford address above.
activity were related to bone size [1]. In the 19th century, considerable scientific interest developed in describing the relationships between bone form and function [2]. Major scientific contributions included the work of Culmann, Meyer, Roux, and Wolff. Roux proposed in 1981 that the apposition and resorption of bone is a biological control process which depends on the local state of stress [2]. Wolff hypothesized that "Every change in t h e . . , function of b o n e . . , is followed by certain definite changes in . . . internal architecture and external confirmation in accordance with mathematical laws" [3 ]. In this paper, the term remodeling will be used to refer to any processes of atrophy, hypertrophy, or change in normal bone turnover that results in a significant alteration of bone shape, size, or microstructure. Although the basic concepts proposed by Roux and Wolff are now universally accepted, the "mathematical laws" relating bone remodeling to bone stresses or strains have yet to be formulated. Furthermore, the mechanical/biological control system or systems that mediate these processes are unknown. In this presentation, a framework is developed in which the relationship between cortical bone tissue mechanical loading history and bone remodeling can be viewed.
Mechanical Demands and Bone Turnover, Hypertrophy, and Atrophy The bones of the body provide a rigid framework for the body tissues and protect internal organs from impact forces. A further mechanical requirement, with considerably more urgency than impact strength, is that the skeleton should not fracture or otherwise fail due to damage caused by repeated strains incurred in normal activities. Repeated loading of bone in everyday activities or prolonged exercise can lead to microscopic damage. The normal osteoclastic and osteoblastic activity (turnover) of
$20
D.R. Carter: MechanicalLoadingHistories IMMOBI~ IZATION
NORMAL ACTIVITY
/
k
SEVERE LOADING
i ~GROWING BONE I
~MATURE BONE
3 STRAIN H/STORY RATE &S ( A E , ~ m N ) At
Fig. 1. Hypothesison the relationshipbetweenbonestrain history rate and the net rate of bone volume change for mature and growing animals. Adapted from Carter [5]. bone serves to repair this damage and maintain the structural integrity of the bone. However, if damage accumulates faster than it can be repaired, fatigue fracture of bone may result. Such fractures have been observed clinically in metatarsals, calcaneous, tibia, fibula, femoral shaft, and femoral neck. Fatigue fractures often occur during prolonged exercise such as marching or long distance running and are especially common in the metatarsals of young military recruits. In many cases, the patient may be unaware of the initiation of a fatigue crack until it is well developed and results in pain or complete fracture. Bone fatigue fracture in vivo is a complex phenomenon in which mechanical damage and biological remodeling and repair processes play an important role. If bone fatigue microdamage accumulates at a slow rate, normal bone turnover may act to repair the damage and maintain the structural integrity of the bone. On the other hand, the creation of microcracks in bone may initiate an osteoclastic response in a manner analogous to that observed in the initial states of fracture healing. The subsequent removal of bone tissue by osteoclasts in a region of bone that continues to be loaded with high cyclic strains will increase those strains and accelerate the accumulation of damage. If the bone hypertrophies at the location of high cyclic strains, those strains will decrease in magnitude, and fatigue fracture will be prevented or delayed. The structural and mechanical demands on the skeleton are very diverse [4]. Some bones, such as the skull bones, have a primary function to protect internal organs from impact loads. Bone tissue at these sites, therefore, experiences minimal repeated strains (or stresses) during normal daily activities. It accumulates negligible mechanical microdamage.
Bones of the lower extremities, on the other hand, usually experience thousands of cycles of loading and unloading every day. The magnitudes, orientations, and sense (tension or compression) of the cyclic principal strains in these bones depend on the location within the bone. There are, therefore, a multitude of different "physiologic" loading histories for bone tissue which display both within bone and among bone variability. From this observation, one must conclude that there is no "optimum" strain history for all bone tissue toward which bone will spontaneously remodel. The strain/remodeling response of bone must be, therefore, either (1) very imprecise, (2) precise but site specific, or (3) both site specific and imprecise. The experience of our research group has evolved a working hypothesis consistent with the idea that the strain remodeling response is both site specific and imprecise. The hypothesis is that the basic structure and mass of each bone is genetically determined by evolutionary pressures and heredity. Normal physiologic activities are required to maintain a level of normal bone mass. Those activities may vary significantly from day to day or month to month without a strong influence on bone mass. There exists, in a sense, a physiologic "band" of activity wherein bone tissue is fairly unresponsive to changes in loading history. This physiologic band is site specific, however. The band of physiologic, acceptable strain histories in the skull is different from that of the anterior midshaft of the femur. That is, a loading history at one site may be entirely normal and would not elicit a remodeling response. That same loading history, if imposed on the bone tissue at another site, may be severely abnormal, however, and result in bone remodeling. Changes in bone mass are affected with increasing intensity as the bone strain history deviates further from the center of the physiologic band. Immobilization can lead to severe bone loss at some skeletal sites. Repeated loading at high strain magnitudes can lead to pronounced hypertrophy. This working hypothesis argues that the strain/ remodeling response is nonlinear (Fig. 1) [5]. One may assume that at a specific site the net rate of the change in bone mass (AB/At) is a function of the time rate of change of a strain history function (AS/At) where At is some time intervals of days or weeks. In a simplistic interpretation, the strain history function, S, can be considered as dependent upon the cyclic strain range (A~), the mean strain (~m), the number of loading cycles (N), and perhaps other characteristics such as loading strain rate or frequency. Two hypothetical curves are presented in Fig. 1 showing the relationship between strain history rate and the rate of bone mass change. The
D. R. Carter: Mechanical Loading Histories curve for mature bone shows that for a fairly broad range of normal activities there will be little stimulus to bone loss or gain. At high levels of strain history rate (severe overload) a hypertrophic stimulus is produced. At very low levels of strain history rate, a pronounced atrophic stimulus is introduced. The curve for growing bone shows that for normal activities there will be a net accumulation of bone mass. The slope of the curve is presented as somewhat greater than that of mature bone, suggesting that growing bone is more sensitive to strain history changes than mature bone. The use of strain rather than stress is based on the personal propensity of the author because of the ease with which the histories can be related to in vitro bone fatigue life and in vivo strain measurements. Many previous investigations of loading-related bone remodeling mechanisms have proceeded on the tacit assumption that a single control system is responsible for maintaining the bone mass, regulating bone loss, and facilitating bone accretion. Treharne [ 1] reviewed several such possible mechanisms which include (1) stress generated electrical potentials in bone, (2) mechanical fatigue microdamage, (3) hydrostatic pressure of the extracellular fluids under load, (4) direct load effects on cell membranes, and (5) changes in mineral solubility under load. Especially in light of the nonlinear control characteristic postulated in Fig. 1, it seems very possible that different mechanisms may be responsible for causing bone atrophy and bone hypertrophy. Two (or more) complementary control systems may work to relate cyclic strains to bone mass regulation. With this perspective, it may be prudent to view bone hypertrophy due to severe loading in a different context from bone atrophy due to immobilization, bed rest, or space flight.
Severe Loading and Bone Hypertrophy Mechanical testing of devitalized bone specimens has demonstrated that, as a material, bone has extremely poor fatigue and fatigue-creep fracture properties [6]. Cyclic strains in the direction of the long bone axis of approximately 0.003 (3,000 microstrain) will fracture small bone specimens after approximately 1 million loading cycles. The bone of the lower extremities will experience between 1 and 10 million loading cycles per year, depending on the activity of the individual. The magnitudes of the cyclic strains depend on location within the bone. The maximum cyclic strains recorded on bone during rigorous physiologic activity in animals of many different species are in the range of 3,000 microstrain, with the compressive strains being somewhat greater
$21 than the tensile strains in long bone diaphyses [79]. These data suggest that at some locations the magnitude of the permissible physiologic cyclic strain is bounded by the fatigue properties of the tissue. The fact that fatigue fractures do occur in man and animals indicates that the "margin of safety" for fatigue fractures is not great. In addition to being a poor material to withstand cyclic loading, the m a n n e r in which bone fatigues is notable. The mechanical data collected during fatigue testing indicates that bone fatigue is accompanied by diffuse structural damage. Specimens tested in the laboratory at strain magnitudes simulating very rigorous activity exhibit an almost immediate loss of bone stiffness and an increase in stress/strain hysteresis [10]. These events indicate the accumulation of bone microdamage. During fatigue loading, internal damage is created in the bone structure which causes a gradual and progressive loss of bone stiffness and strength prior to major, detectable crack formation. The damage accumulation, as evidenced by mechanical events, is related to specific structural defects created in the bone structure [7]. In bone tissue that accumulates tensile damage, there is extensive failure at the osteon cement lines and interlamellar cement bands. The nature of this damage is such that direct insult to bone cells and canaliculi is minimal. In bone tissue that accumulates compressive fatigue damage, numerous oblique microcracks are created. Direct insult to bone ceils and canaliculi is extensive and, indeed, the formation of microcracks appears to be influenced to some extent by the stress concentrations created by vascular canals and lacunae. This observation is important when trying to relate the fatigue behavior of devitalized bone (which reflects only mechanical phenomena) to the in vivo fatigue situation that reflects both mechanical and biological processes. The more extensive cellular insult to bone cells during compressive fatigue loading may provide a greater stimulus for biological repair of microdamage and extend the in vivo fatigue life of bone. Chamay and Tschantz [11] examined canine ulnae in an investigation of the stress-related remodeling response in adult dogs. In one group of dogs, a small segment of the radius was resected and the animals were allowed to walk on their weakened forepaws, the load being transmitted only through the ulna. This procedure resulted in either fatigue fracture or in massive bone hypertrophy, apparently depending upon the activity of the dog and inherent strength of the ulna. Histologic examination after resection of the radius showed oblique lesions on the concave ulna cortex (a region of high compressive stresses) several hours after activity was resumed.
$22 These lesions were similar to those formed during compressive fatigue loading of devitalized bone specimens [6]. The remodeling process occurred over a 9-month period and resulted in a considerably enlarged diaphysis. Chamay and Tschantz contended that the cellular insult resulting from the oblique lesions may have served as a remodeling stimulus. However, they noted that hypertrophy also occurred in regions where lesions could not be seen. Animal experimentation and clinical experience certainly suggest that the accumulation of mechanical microdamage can serve as a stimulus to bone hypertrophy in the case of significantly increased physical activity. However, the creation of microdamage in normal adult bone is minimal at cyclic strain levels less than 2,000 microstrain. If bone can hypertrophy in response to strain changes below this level, it is likely that there is another bone hypertrophy control mechanism acting such as streaming potentials, cell pressure, or direct load effects on cell membranes. If such additional hypertrophic control systems exist, they must work in conjunction with mechanical microdamage stimuli to affect an integrated response of bone to above normal mechanical demands. Clinical investigations have shown that rigorous physical activities can result in above average bone mass [5]. These studies, however, have tended to be focused on individuals who engage in very rigorous activities and have begun these activities at an early stage of skeletal maturity. As suggested in Fig. 1, immature bone may have a greater ability to respond to such loading than more mature bone. Woo [12] reported that exercise training for 12 months caused a significant increase in femoral cortical thickness in young pigs. This hypertrophy resuited in increased structural strength, although the bone material properties were unchanged. Based on these results and data from previous "stress shielding" experiments, Woo hypothesized that bone deposition under moderate physiologic loading activities was linearly related to activity levels but immobilization could cause drastic, disproportionate losses in bone mass [ 12]. This view is similar to that shown for the immature bone curve (Fig. 1) for loading below the "severe overload" regime. The response of mature bone to moderate increases in strain history is at this point, controversial. Most animal models that have been used to study the influence of small or moderate increases in load have necessitated the imposition of trauma and disruption of the bone blood supply at or near the bone under investigation. This situation can, and does disrupt normal bone physiology and cause bone formation. Several researchers have been able to demonstrate bone deposition in response to a surgical procedure
D.R. Carter: MechanicalLoadingHistories designed to increase cyclic strain magnitudes to levels not expected to result in mechanical microdamage [ 13-15 ]. The influence of surgical trauma near the sampling site on bone deposition in these models, however, is difficult to assess. Carter et al. [8] conducted experimental studies of adult canine radii, to determine how increased cyclic strain levels affect bone remodeling. Alterations of cyclic strain patterns in the midshaft of the radius were produced by unilateral resection of a portion of the distal ulna. A distal resection was used to avoid surgical trauma near the mid-radius sampling site. The midshaft of the contralateral radius served as a control. After the operation, the dogs were subjected to a controlled exercise program so that the total number of loading cycles (approximately 20,000) over an 8 week period could be estimated. The dogs were then sacrificed and the differences in the radius midshafts were assessed histologically and mechanically. To estimate the alteration of cyclic strain levels created by the operations, other dogs were instrumented bilaterally with three strain gages bonded at different circumferential locations on the midshafts of each radius. The cyclic strain patterns on both the control and the experimental limb were recorded during normal gait. Intracortical distributions of cyclic strains were determined using mathematical models in conjunction with the recorded strain values. In our dog model, we attempted to impose known increases in the strain histories of bone tissue. The magnitude of the cyclic strain amplitudes were markedly increased by more than a factor of two. The in vivo strain histories imposed on the experimental radii were not, however, severe enough to produce fatigue failure of the tissue. The bone remodeling response illicited by this increased strain history over an eight week period was undetectable in 3 dogs and subtle in the fourth dog. These findings indicate that a massive stimulus to bone turnover was not induced. It may be that a mild stimulation of bone formation was produced in all of the dogs but the 8 week experimental period was not long enough to result in substantial remodeling differences. The increased loading history imposed in our canine experiment can be viewed in a different perspective when placed in the context of the physiologic strains encountered in normal activities. All of our dogs were restricted in their activities since they were not allowed to run about freely. The strain histories in the control limb were therefore restricted to those which would be caused by normal walking. The strain histories imposed on the experiment radii were significantly increased when compared with the contralateral controls. The strain histories in the
D. R. Carter: Mechanical Loading Histories experimental radius m a y not have been increased, however, when compared with those experienced by an active, freely running dog, in a normal domestic environment [9]. It can be hypothesized that the experimental and control limbs were, in fact, exposed to strain histories within the "normal activity" band of Fig. 1.
Strain Shielding and Bone Atrophy At low levels of cyclic bone strain there is minimal mechanical microdamage to the bone tissue. It is difficult to imagine that a further decrease in the rate of this damage accumulation could trigger the bone loss observed in immobilized bone. It is likely, therefore, that other control systems are active in bone mass regulation under these circumstances. Although there has been a considerable amount of clinical and experimental work concerning immobilization-induced osteopenia, there are no quantitative data available relating bone atrophy to strain changes in these studies. The lack of bone stress or strain quantification in this area of research is primarily due to the complexity of theoretical modeling of the musculoskeletal systems involved. One method of reducing the magnitude of the cyclic strains in bone tissue is the application of a fracture plate. The plate transmits a substantial portion of the loads that are normally transmitted by the bone. Extensive bone remodeling and a net bone loss may occur which many researchers attribute to the changes in in vivo bone stresses [5, 12, 16]. Additionally or alternatively, bone loss and remodeling may be influenced by the vascular insult introduced by the fracture a n d / o r the presence of the plate [17]. Experimental studies indicate that, in general, greater bone remodeling and bone loss is observed when stiff plates are used than when more flexible plates are used. Carter et al, [16] attempted to assess the alterations in the in vivo cyclic bone stresses due to the application of various plates on the canine femoral shaft. In their analyses, Carter et al, used the in vivo strain gage results reported in their earlier study [ 18] to estimate the in vivo stress field in the normal canine femoral midshaft during the stance phase of the gait cycle. The plate configurations analyzed were those used by previous investigators when studying the influence of plating on bone remodeling. The magnitude of the reduction in the loads borne by the bone tissue and the degree of shift in the bone stress neutral axis during the stance phase of gait was influenced by the geometry of the plate, the plate elastic modulus, and the location of plate application. From a correlation of the calculated
$23 alterations in bone stresses with the resulting measured changes in bone mass, it appears that bone remodeling is very sensitive to small changes in cyclic bone stresses. Changes in cyclic bone stresses of 1 M P a (less than 1% of the ultimate strength) can cause measurable differences in bone atrophy after a period of a few months. The apparent sensitivity of cortical bone resorption in response to a reduction in cyclic strain magnitudes was also recognized by Woo [12].
Conclusion This paper presents a conceptual framework in which the adaptive remodeling response of cortical bone can be viewed and investigated. At this point, most of the concepts are presented in qualitative terms. The greater use of in vivo strain instrumentation and analysis techniques, however, provides hope that more quantitative relationships can be derived. I believe the major questions that should be addressed in future work are (1) To what extent is the remodeling response site specific? (2) Is the response really nonlinear as depicted in Fig. 1 ? (3) To what aspects of the loading history does bone respond? (4) W h a t control systems related to loading history can be isolated to produce a bone remodeling response? (5) W h a t is the effect of age on the remodeling response? If the response is better characterized phenomenologically, the cellular and biochemical aspects can be investigated with greater focus and direction.
References 1. Treharne RW (1981 ) Reviewof Wolff's Law and its proposed means of operation. Orthopaedic Review 10:35-47 2. Roesler H (1981) Some historical remarks on the theory of cancellous bone structure (Wolff's Law). In: Cowin SC (ed) Mechanical properties of bone. American Society of Mechanical Engineers Publication AMD-Vol 45:27-42 3. Wolff J (1982) Das gesetz der transformation der knochen. A. Hirshwald, Berlin 4. Currey JD (1983) Mechanical and biological requirements for structural competencein the skeleton(what shouldbones be designedto do?) Presented at Kroc FoundationConference on FunctionalAdaptation in BoneTissue, Santa Ynez Valley, CA, July 11-15 5. Carter DR (1982) The relationships between in vivo bone stresses and cortical bone remodeling. In: Bourne JR (ed) CRC Critical Reviews in Bioengineering, Boca Raton, FL, CRC Press, 8:1-28 6. Carter DR, Caler WE (1983) Cycle-dependentand time-dependent bone fracture with repeated loading. J Biomech Eng 105:166-170
$24 7. Carter DR, Caler WE, Spengler DM, Frankel VH (1981) Fatigue behavior of adult cortical bone--the influence of mean strain and strain range. Acta Orthop Scand 52:481-490 8. Carter DR, Harris WH, Vasu R, Caler WE (1981) The mechanical and biological response of cortical bone to in vivo strain histories. In: Cowin SC (ed) Mechanical properties of bone. American Society of Mechanical Engineers Publication AMD-Vol 45-81-92 9. Rubin CT, Lanyon LE (1982) Limb mechanics as a function of speed and gait: a study of functional strains in the radius and tibia of horse and dog. J Exp Biol 101:187-211 10. Carter DR, Hayes WC (1977) Compact bone fatigue damage. I. Residual strength and stiffness. J Biomech 10:325-338 11. Chamay A, Tschantz P (1972) Mechanical influences in bone remodeling. Experimental research on Wolff's Law. J Biomech 5:173-180 12. Woo S.L-Y (1981) The relationships of changes in stress levels on long bone remodelling. In Cowin SC (ed) Mechanical properties of bone. American Society of Mechanical Engineers Publication AMD-Vol 45:107-129
D . R . Carter: Mechanical Loading Histories 13. Liskova M, Hert J (1971) Reaction of bone to mechanical stimuli. Folia Morphologica 19:301-317 14. Lanyon LE (1981) The measurement and biological significance of bone strain in vivo. In: Cowin SC (ed) Mechanical properties of bone. American Society of Mechanical Engineers Publication AMD-Vol 45:93-106 15. Churches AE, Howlett CR (1981) The response of mature cortical bone to time-varying loading. In: Cowin SC (ed) Mechanical properties of bone. American Society of Mechanical Engineers Publication AMD-Vol 45:69-80 16. Carter DR, Vasu R, Harris WH (1981) The plated canine femur--relationships between the changes in bone stresses and bone loss. Acta Orthop Scand 52:241-248 17. Gunst MA (1980) Interference with bone blood supply through plating of intact bone. In: Uhthoff HK (ed) Current concepts of internal fixation of fractures. Springer-Verlag, Berlin, pp 268-276 18. Carter DR, Vasu R, Spengler DM, Dueland RT ( 1981 ) Stress fields in the unplated and plated canine femur calculated from in vivo strain measurements. J Biomech 14:63-70
